Available online at www.sciencedirect.

com

ScienceDirect
Journal of Electrical Systems and Information Technology 5 (2018) 144–157

Vibration analysis including stator, rotor, housing and dynamic

response analysis of Flux Reversal Generator

B. Vidhya a,∗ , K.N. Srinivas b

a Department of Electrical and Electronics Engineering, B.S. Abdur Rahman Crescent Institute of Science and technology, Chennai 600048, India
b Department of Electrical and Electronics Engineering, SRM Institute of Science and Technology, Ramapuram, Chennai 89, India

Received 26 August 2015; received in revised form 12 April 2017; accepted 3 February 2018
Available online 20 February 2018

Abstract
This paper attempts to present numerical solution for the vibration analysis of a Flux Reversal Generator (FRG). A three-
dimensional finite-element modal analysis (3-D FEA) methodology has been used for this purpose. Vibrations induced due to
mechanical and unbalanced reasons have been analyzed. The 3-D modal analysis undertaken identifies the vibrations caused in
every part of FRG (stator, shaft, end shields, bearings, and housing), which has been systematically documented. Modal frequencies
and the respective speeds which are to be skipped immediately in order to operate the machine always in a safer acoustic zone have
been identified and tabulated. To fine tune this, a harmonic analysis has been performed. Up to 10,000 Hz has been considered and
frequencies (and proportional speeds) such as 1500 Hz, 1488.62 Hz and 1568.78 have identified as the speeds to be avoided. Apart
from frequently reported modal analysis on electrical machines, this 3-D vibration analysis for FRG considering every part of the
machine will be helpful to examine the vibration in the machine as a whole. This forms the claim of the paper.
© 2018 Production and hosting by Elsevier B.V. on behalf of Electronics Research Institute (ERI). This is an open access article
under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Flux Reversal Generator; Modal analysis; Finite element method; Harmonic response; Frequency domain analysis

1. Introduction

1.1. Flux reversal machines (FRM)

FR machine is developed by combining the advantages of SRM (simple construction, fault tolerance, mechanical
robustness) and DSPMM (high energy magnets placed in a stator back iron). It is a doubly salient machine. It has a
stator with even number of projecting poles. Each stator pole has a concentrated field winding on its pole body and

∗ Corresponding author.
E-mail addresses: vidhyabheeman@gmail.com (V. B.), knsrinivas1967@gmail.com (S. K.N.).
Peer review under the responsibility of Electronics Research Institute (ERI).

https://doi.org/10.1016/j.jesit.2018.02.004
2314-7172/© 2018 Production and hosting by Elsevier B.V. on behalf of Electronics Research Institute (ERI). This is an open access article under
the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 V. B., S. K.N. / Journal of Electrical Systems and Information Technology 5 (2018) 144–157 145

Fig. 1. Operation of flux reversal machine: (a and c) flux does not link field winding. (b and d) Flux has a closed path through the yoke and has flux
linkage.

oppositely polarized permanent magnets (PM) on its pole face. Rotor of the machine too has even number of projecting
poles, but without any kind windings on it. The principle of operation of FRG is simple. On the stator pole-rotor pole
aligned position there is no flux linkage with the stator field winding, as the flux set up circulates completely within
each stator pole around PMs. This is the equilibrium position in which there is no flux linkage (or MMF). When the
rotor pole overlaps other magnets, after moving certain amount of degrees of rotation, the flux will start to pass through
the phase coil, ultimately reaching maximum/minimum. In these positions, flux linkage flux linkage (or MMF) occur.
This produces the useful torque for the FR machine (Deodhar et al., 1997).
This is depicted in Fig. 1. Fig. 1(a) and (c) shows two equilibrium positions in the course of the rotor’s rotation
in CCW direction. There will be no MMF in these positions. Fig. 1(b) shows the displacement of rotor after 30◦ and
Fig. 1(d) shows the same after a further 30◦ + 30◦ , that is 60◦ displacement of the rotor. These are the rotor positions at
146 V. B., S. K.N. / Journal of Electrical Systems and Information Technology 5 (2018) 144–157

which there is MMF and flux linkage and hence sinusoidal EMF is induced, respectively reaching positive maximum
and negative maximum.
Apart from the useful torque (or EMF) produced, there are always two problems associated with FR machines, and
they are cogging and critical speeds. This paper addresses these problems.

1.2. Cogging in FRM

However, electric machines involving PMs exhibit a severe Cogging torque (Dorrel et al., 2003). This is a torque
which gets produced due to the interaction between the PMs of the stator and the moving rotor of the machine. This
torque is position dependent and it has periodicity per revolution, in the sense that it repeats in every successive 360◦
of rotation at the same instances. This is also called as a zero torque and so a jump from useful torque and zero torque
always keeps happening in every 360◦ of rotation at the cogging instances. Thus, cogging torque is an undesirable
component for the operation of motors, for two reasons: (i) objectionable jumps will hamper the smooth operation,
and (ii) the associated disturbing vibration and acoustic noise will cause annoyance.
At high speeds, as the motor moment of inertia of the machine and the rotor’s fast ROM filter out the effect of cogging
torque, its effect will be negligible. But it cannot be done so at lower speeds, because the symptom of jerkiness will be
severe and notable So, at lower and moderate speeds, cogging effect will be high. As the cogging forces rotate with
the rotor, they cause vibrations at the frequency of rotation. When the frequency of the exciting force coincides with
one of the natural frequencies of the body, the condition of resonance occurs. When machine is operated in resonance
large vibration will occur, resulting in premature fatigue failure, damages the components such as bearings, and gears,
and even causes damage to the internal organs of us.
Attempts are researched for the reduction of cogging torque (Kim et al., 2005). This issue is not only addressed
for FR machines, but has been addressed for other electric machines (Srinivas and Arumugam, 2005) also. In this
focus, this paper presents a systematic documentation of 3-dimensional numerical analysis to identify the modal
frequencies which are contributing for the production of cogging torque and the associated vibrations in FR machines.
This addition in the FR machine literature, will help the practicing engineers to have a guideline paper to undertake
vibration analysis in FR machines and identify the noise producing modal frequencies, which when avoided will help
to circumvent cogging torque.

1.3. Critical speed

Cogging is highly pronounceable in low speed operations and when the machine is used for high speed applications
the issue of critical speeds of the generator is to be addressed (Krishnan, 2010; Lawrenson, 1965). In rotor dynamics,
the critical speed is the theoretical angular velocity that excites the natural frequency of the structure in dynamic
operation. The structure does not only include rotor, but also includes other stationary parts vulnerable for vibration
such as stator, shaft, yoke, housing or platform. As the speed of rotation approaches the object’s natural frequency, the
object will begin to resonate, which is dangerous, in the sense that it will dramatically increase the vibration of the
structure. The resulting resonance occurs regardless of orientation. When the rotational speed is equal to the numerical
value of the natural vibration, then that speed is referred to as critical speed. Thus identifying critical speed also is a
prime important.
If the rotor is rotating above the critical speed it causes vibration which leads to the failure of generator parts. To
enhance the quality of the machine characteristics and to avoid fatigue failure, this machine has to be designed with
low frequency and low noise level. Hence the critical speed or natural frequency of the Flux Reversal Generator (FRG)
is to be calculated. To find the natural frequency, modal analysis is required. This can be done using the finite element
methodology, which is proved in other fields of engineering also (Valipour et al., 2017; Valipour, 2016, 2017).

1.4. Scope of the paper

With the above points in mind, the modal analysis and the identification of critical speeds for FRG have been
presented in this technical paper. This vibration analysis is done for the whole generator, part-by-part, including stator,
rotor, shaft, end shields, bearing and foundation. Thus quantitative information about the vibration of the complete
structure will be in hand. Finally, the natural frequency (for critical speed calculations) and mode shapes of the
 V. B., S. K.N. / Journal of Electrical Systems and Information Technology 5 (2018) 144–157 147

Table 1
Material properties.
Material properties Stator Rotor Magnet Windings

Young’s modulus (N/m2 ) 12 × 1011 12 × 1011 1.6 × 1011 117 × 109

Mass density (kg/m3 ) 7750 7750 7500 8940
Poisson’s ratio 0.3 0.3 0.24 0.335

FRG are calculated and the vibration response of the generator is analyzed by the frequency domain response. This
comprehensive report fills the gap in FRG literature, as far as the vibration analyses are considered.

2. Stages of finite element (FE) modelling for vibration analysis of FRG

The cross section of the FRG under investigation modeled in CAD is shown in Fig. 2.
The CAD model is developed for a three phase 6/8 pole small scale Flux Reversal Generator of 450 W. The FE
modeling has three major stages and they are: pre-processing, processing and post-processing. The first step is to model
the main solid parts of the generator: stator, rotor, shaft, end shields, bearings and frame, which form the pre-processing
stage. The stator, rotor assembly is first modeled in 2D and extended up to the stack length of 40 mm to form the 3D
model. The shaft has a projecting length of 28 mm. The end rings are tightly mounted on both sides of the FRG to
support the rotor assembly. The thickness of the each bearing on the shaft is 5.5 mm. This structure is modeled with
element type solid 45 in ANSYS. This has 3◦ of freedom, namely Ux , Uy , Uz . The 3D model of the FRG is shown in
Fig. 3.
In the processing stage the material properties, such as material used, Young’s modulus, Poisson’s ratio, mass density
resistivity of every part of the structure must be declared at the respective parts (Miller and McGilip, 1987). In this
case, the properties of PMs have to set as a special case. The major mechanical properties used for vibration analysis
are given in Table 1. These are in-built in the software material library.
However, certain parts will not be in-built and hence they are to be modeled discretely as per the requirement. In this
FRG’s case, the mechanical coupling is to be modeled separately as there is no provision for in-built selection of this in
the software library. The procedure coined is called nodal coupling. The nodal couplings are applied between the shaft
and bearing end cover at the location of the bearing. The nodes are coupled in Ux , Uy , Uz direction to represent the
bearings. Coupling is a way to force a set of nodes to have the same DOF (Degree Of Freedom) value. The command
used for coupling is CPINTF.
The foot of the generator is clamped to represent the boundary conditions, as shown in Fig. 4. The axial symmetry
boundary condition is applied. This will solve half of the model, which solution will form the similar solution for the
other half of the model also, thus reducing the time and computational efforts.
The next step in the pre-processing stage of FE process is to generate the mesh of each solid part. This is called as
meshing and it is a crucial step in design analysis. It subdivides the model into small pieces of elements connected at
common nodes. FE analysis predicts the behavior of the model by combining the information obtained from all small
elements making up the model. Meshing completes the preprocessing stage of any FE method. The meshed model of
the various parts of FRG (rotor, end shield, enclosure and footage) is shown in Fig. 5.
Once the finite element meshed model is generated, the next step is to conduct the analysis. This is the processing
stage. In this work, modal analysis using the modal stress solver forms the processing. The modal solver allows the
evaluation of the natural frequency and mode shapes of the structures. The identification of resonant frequency that
could create severe vibration and noise is also mainly evaluated. Fig. 6 shows the general procedure adopted in FE
methodology of vibration analysis.

3. Modal analysis

The intelligent usage of the results obtained from FE analysis forms the post-processing. The governing equation
of motion of the free vibration without considering damping can be written as,
148 V. B., S. K.N. / Journal of Electrical Systems and Information Technology 5 (2018) 144–157

Fig. 2. Dimensions and cross section of Flux Reversal Generator. (a) Radial cross section (b) axial cross section.
 V. B., S. K.N. / Journal of Electrical Systems and Information Technology 5 (2018) 144–157 149

Fig. 3. 3-D model of FRG. (a) 3-D model of FRG, showing stator, rotor and end shield. (b) 3-D model of FRG with shaft, end cover, footage with
clamping.

Fig. 4. Bearing model: (a) nodal coupling (b) footage clamping to set boundary condition. The Ux , Uy ,Uz direction can be seen.
150 V. B., S. K.N. / Journal of Electrical Systems and Information Technology 5 (2018) 144–157

Fig. 5. Meshed model of FRG. (a) Rotor, (b) stator and magnet with footage and (c) whole model of FRG with end shield and housing.

Fig. 6. Steps involved in vibration analysis.

 V. B., S. K.N. / Journal of Electrical Systems and Information Technology 5 (2018) 144–157 151

Table 2
Natural frequency of FRG.
Mode. no Mode Location Frequency (Hz) Speed (RPM)

1 Horizontal bending Housing 1488.62 11,164

2 Axial bending 1568.78 11,766
3 Radial bending 4119.94 30,899
4 Rotor bending Rotor 3372.43 25,293
5 Shaft bending Shaft end 3848.85 28,866

M Ü + KU = 0 (1)
where M and K are mass and stiffness matrices. Ü, U are the vectors of generalized displacement and acceleration.
The solution to the homogeneous Eq. (1) is given by,
U = Øi Sin (ωi t + θi ) (2)
with i − 1, 2, 3 (3 degree of freedom).
In this equation, Øi is a vector of order n, ωi is the angular frequency of mode i and θ i denotes the phase angle.
When Eq. (2) is differentiated twice, we obtain,

Ü = − ωi2 Øi Sin (ωi + θi ) (3)

 
−Mωi2 + K Øi sin (ωi t + θi ) = 0 (4)

and since sin (ωi t + θi ) = 0 for most times, we write,

 
−Mωi2 + K Øi = 0 (5)

Eq. (6) yields eigen solutions,

 2

ω12 Ø1 , ω2 Ø2,......................... ωn2 Øn (6)

The three dimensional governing equation to find the modal frequencies is,
∂2 Ø ∂2 Ø ∂2 Ø
+ 2 + 2 + ω2 Ø = 0 (7)
∂x2 ∂y ∂z
This equation is solved for Ǿ, the order of vibration, using FE numerical approach.
This non-linear modeling finds application in other disciplines also (Valipour, 2012a,b).
Fig. 7 shows the vibration modes of the machine housing with foundation. The vibration spreads fully over the
housing in horizontal direction (Fig. 7(a)), axial direction (Fig. 7(b)) and radial direction (Fig. 7(c)) at modal frequencies
of 1488.62 Hz, 1568.78 Hz, 4119.94 Hz respectively. The natural frequency of this mode is observed to be not severe.
Consider Fig. 8, which is the results of simulation of alarming vibration zones. At a modal frequency of 3372.43 Hz,
the rotor and shaft undergo bending and thus arresting the housing, which produces high acoustic noise. This is shown
in Fig. 8(a). Fig. 8(b) shows a severe shaft deformation at the modal frequency of 3848.85 Hz. It is observed that shaft
end vibration does not spread to rotor or housing. From the modal analysis, natural frequencies are found using which
the critical speeds are obtained (using 60 f/p), which will indicate that, the generator should not be operated at such
speeds. These speeds are tabulated in Table 2. The other higher order frequencies are neglected that is given in Table 3.

4. Frequency response

Modal frequencies indicate the frequencies (or proportional speeds) at which the machine shall not function. Critical
speed indicates the same at high speeds. These two were predicted for the FR machine in the above two sections.
Harmonic analysis, presented in this section, is the third and final sort of vibration analysis on structures. A harmonic,
or frequency-response, analysis is used to determine the response of the structure under a steady-state sinusoidal (that
152 V. B., S. K.N. / Journal of Electrical Systems and Information Technology 5 (2018) 144–157

Fig. 7. Housing vibration. (a) Mode frequency: 1488.62 Hz, horizontal bending (b) mode frequency: 1568.78 Hz, axial bending and (c) mode
frequency: 4119.94 radial bending.
 V. B., S. K.N. / Journal of Electrical Systems and Information Technology 5 (2018) 144–157 153

Fig. 8. Modal shape at 3372.43 Hz, rotor and shaft bending. (b) Modal shape at 3848.85 Hz shaft bending.

is the harmonic) loading at a given frequency. It used to determine the steady-state response of FRG to loads that
vary harmonically with time. Thus, the aim of frequency response is to obtain the FRM structure response at various
frequencies with respect to displacement.
In the FEA tool used by the authors,‘harmonic analysis tool’ is selected to analyse the harmonic response of any
structure (here, FRG).
It is worth to note that prior to the harmonic analysis run, the free vibration analysis in time domain (above two
sections) should be completed. Then only the frequency range in which the harmonic analysis to be performed can be
identified.
The unbalanced force is given by, FU = m.ω2 .e where m = mass of the rotor (3.23/9.81=0.33 kg) and e is eccentricity
of the rotor given by e = Vω , where V is the velocity in mm/s and ω is the angular speed (rad/s). Under the operating
speeds of 9000 RPM corresponding to balancing grade Q 2.5 (will vibrate at 2.5 mm/s), the unbalanced force Fu is
154 V. B., S. K.N. / Journal of Electrical Systems and Information Technology 5 (2018) 144–157

Table 3
Modal analysis result.
Set Time/frequency Load step Sub step Cumulative

1 1488.6 1 1 1
2 1568.8 1 2 2
3 3372.4 1 3 3
4 3848.8 1 4 4
5 4119.9 1 5 5
6 4319.1 1 6 6
7 5522.1 1 7 7
8 6170.8 1 8 8
9 6332.9 1 9 9
10 9312.1 1 10 10
11 10,267 1 11 11
12 11,408 1 12 12
13 11,495 1 13 13
14 12,689 1 14 14
15 14,283 1 15 15
16 14,674 1 16 16
17 14,997 1 17 17
18 15,471 1 18 18
19 15,902 1 19 19
20 16,324 1 20 20

computed as 0.76 N. This force is applied to the centre node of the stator. The arrow Mark at the center of the stator is
shown in Fig. 9.
This constant force is applied over a frequency range of 1 Hz to 10,000 Hz. The simulation results of harmonic
response of the stator and rotor are shown in Fig. 10. Fig. 10(a) and (b) shows that large displacement occurs cor-
responding to the frequency of 1500 Hz. Since 1488.62 Hz and 1568.78 Hz are predominant, this corresponds to the
modes 1 and 2 respectively as shown in the Table 2.
The critical speeds are identified as 11,164 rpm and 11,766 rpm. Hence it is inferred that, the FRG should not operate
at these speeds.
It can be observed that the rotor and stator eccentricity reaches a maximum of 1.4 ␮m. As this eccentricity is of
negligible micrometers, it is concluded that the rotor dynamics of the considered FRG is in acceptable limits.

5. Conclusion

A 3-D modal analysis to study the vibration characteristics of FRG has been addressed in this paper. The stator,
stator frame, rotor, end shield, bearings, shaft, and foundation have been modeled in three dimensions to study the
vibration in FRG as a whole. Certain parts of the machine, such as bearings which are not in-built, are separately
modeled. A step-by-step 3-D vibration analysis procedure for modal analysis, critical speed identification and harmonic
analysis are documented for a 6/8 FRG. Natural frequency and mode shapes of the generator have been calculated
by mean of the FE analysis. From the results, it could be seen that the bending mode frequency is 3848.85 Hz,
which causes high vibration and noise. Harmonic response could be performed in the frequency domain based on
the calculated frequencies. The large displacement can occur when centrifugal force frequencies are near the natural
frequency of modes 1 and 2. The procedure can be adopted for any pole combination and dimensions of FRG to
thoroughly perform vibration study in three dimensions. This will help to fine tune the design to be declared for end
product.
 V. B., S. K.N. / Journal of Electrical Systems and Information Technology 5 (2018) 144–157 155

Fig. 9. (a) Application of unbalanced force at stator for harmonic analysis. (b) The screen of frequency response analysis.
156 V. B., S. K.N. / Journal of Electrical Systems and Information Technology 5 (2018) 144–157

Fig. 10. Frequency domain response (a) stator (b) rotor.

References

Deodhar, R.P., Andersson, S., Boldea, I., Miller, T.J.E., 1997. Flux reversal machine: a new brushless doubly-salient permanent-magnet machine.
IEEE Trans. Ind. Appl. 33 (4), 925–934.
Dorrel, D.G., Chindurzha, I., Butt, F., 2003. Operation, theory and comparison of flux reversal machine—is it a viable option? Proc. Int. Conf. Power
Electron. Drives., 253–258.
Kim, Tae Heoung, Won, Sung Hong, Bong, Ki, Lee, Ju, 2005. Reduction of cogging torque in flux reversal machine by rotor teeth pairing. IEEE
Trans. Magn. 41 (October (10)), 3964–3966.
Krishnan, R., 2010. Permanent Magnet Synchronous and Brushless DC Motor Drives. CRC Press.
Lawrenson, P.J., 1965. Development of application of reluctance motor. Proc. Electron. Power (June), 195–198.
Miller, T.J.E., McGilip, M., 1987. PC CAD for switched reluctance drives. In: Proc. IEE Electric Machines and Drives Conf., London, U.K.,
December, pp. 360–366.
Srinivas, K.N., Arumugam, R., 2005. Analysis and characterization of switched reluctance motors: part II—flow, thermal and vibration analyses.
IEEE Trans. Magn. 41 (March (3)), 1321–1332.
Valipour, Mohammad, Gholami-Sefidkouhi, Mohammad Ali, Raeini-Sarjaz, Mahmoud, 2017. Selecting the best model to estimate potential
evapotranspiration with respect to climate change and magnitudes of extreme events. Agric. Water Manag. 180 (January (Part A)), 50–60.
 V. B., S. K.N. / Journal of Electrical Systems and Information Technology 5 (2018) 144–157 157

Valipour, Mohammad, 2012a. A comparison between horizontal and vertical drainage systems (include pipe drainage, open ditch drainage, and
pumped wells) in anisotropic soils. IOSR J. Mech. Civil Eng. (IOSR-JMCE) 4 (1).
Valipour, Mohammad, 2012b. Ability of Box–Jenkins models to estimate of reference potential evapotranspiration (a case study: Mehrabad Synoptic
Station, Tehran, Iran). IOSR J. Agric. Vet. Sci. 5 (1).
Valipour, Mohammad, 2016. How much meteorological information is necessary to achieve reliable accuracy for rainfall estimations? Agriculture
6.
Valipour, Mohammad, 2017. Analysis of potential evapotranspiration using limited weather data. Appl. Water Sci. 7 (March (1)), 187–197.